a comparison of two generic trap types for monitoring ...

Journal of the American Mosquito Control A.ssociation. 25(I):58-65. 2009 Copyright © 2009 by The American Mosquito Control Association, Inc.

A COMPARISON OF TWO GENERIC TRAP TYPES FOR MONITORING MOSQUITOES THROUGH AN ANNUAL CYCLE IN TROPICAL AUSTRALIA LUBOMIR BISEVAC,' DONALD C. FRANKLIN,' GRANT J, WILLIAMSON'- AND PETER I, WHELAN' ABSTRACT. We compare the community composition, abundance, and seasonality of mosquito species detected by the encephalitis virus surveillance (EVS) CO2 traps and Centers for Disease Control and Prevention (CDC) light traps. Traps were run concurienily for a year during routine weekly monitoring in the vicinity of the city of Darwin in northern Australia. The EVS CO2 traps detected far more individuals than CDC light traps notwithstanding a weaker suction fan, but species richness was similar. Regardless of variation in community composition among sites, differences between trap types were remarkably consistent. Seasonal trends in the abundance of 5 key species from each trap type were similar, but markedly more so in strongly seasonal species. Although EVS CO:, faps outperformed CDC light traps for routine monitoring, the historical transition from the latter to the former is unlikely to have major consequences lor the identifÈCation of community composition or detection of seasonal trends in key species. KEY WORDS

Attractant, tropical, mosquito control, vector, swamp

INTRODUCTION Monitoring mosquito populations is a key responsibility of health ageneies that facilitates the identification of disease risk and efficient spatial and temporal targeting of control measures. Monitoring of niosquitoes requires the use of standard trap and bait types that permit the detection and enumeration of the range of species of concern (Service 1993). Although it is possible to employ a multiplicity of trap designs and attractants to ensure that all possible .species are adequately detected (Van Essen et al. 1994. Russell 2004. Muturi ct al. 2007), time and resource constraints often favor a simple, generic monitoring system. The choice of such a monitoring system, and knowledge of its ability to consistently detect the range of species of interest, is critical to program success. Amongst the range of possible attractants, light and CO2 are amongst the most generic (Service 1993). Historically, light traps have been the primary tool for monitoring mosquito populations (Sudia and Chamberlain 1962, Odetoyinbo 1969, Service 1970), but these have now largely been superseded by traps baited with CO2. which may be provided in the form of dry ice. which emits a plume of CO2 over time (Gillies 1980. Russell and Whelan 1986, Cooperband and Cardé 2006, Bishop et al. 2008). However, light traps are still used in many settings, especially where dry ice is not readily available.

Light traps with CO2 as a supplementary bait catch more mosquitoes than light traps alone (Reisen et al. 2000, Russell 2004, Muturi et al. 2007). but most evaluations of trap types and baits consider only a few key species over relatively short experimental periods. In this study, we report on a 1-yr segment of a longterm monitoring program for the tropical city of Darwin in northern Australia, The study year comprised the overlap period between 2 forms of monitoring—the transition from customized Centers for Disease Control and Prevention (CDC) light to encephalitis virus surveillance (EVS) CO2 traps—in which 6 matched pairs of traps were evaluated weekly. Beyond the issue of the attractiveness of bait type to dilTerent species, EVS traps have the practical advantage of being smaller, lighter, and easier to power, and of attracting markedly fewer non-target species. Our aims were to compare tbe efTicacy of tbese trap types in a monitoring program in 1) detecting and measuring the abundance of all species, 2) identifying community structure independent of differences among sites, and 3) identifying seasonal patterns of abundance for 5 key management species. The key species were Culex annulirostris Skuse, Aeäes vigilax (Skuse), Anopheles haneroftii Giles. Anopheles farauti (Laveran), and Coquiltettidia xanihogasler (Edwards). MATERIALS AND METHODS Study area

' School for Environmental Research. Charles Ditrwin (Jniversity, Diirwin. NT 0909. Australia. " School of Plant Science, University of Tasmania, Private Bag 55. Hobart, TAS 7001. Australia, ' Mcdicítl Entomology. Centre for Disease Control, NT Department of Health and Eamilies, Darwin, NT 0811, Australia.

Darwin (12'^27'S, 13O°5O'E) is situated on the tropical north coast of Australia ( Fig. 1 ), The city and its immediate surrounds are flat to gently undulating and below 50 m above sea level. An extensive system of coastal mangroves, salt marshes, and freshwater wetlands lies immediate-

MARCH 2009

0 0.5 1

TRAP TYPE COMPARISON

I Kilometers

59

Quickbird Image (c) 2004 Digital Globe

Fig. I. The city ol" Darwin. Northern Territory, Australia, with locations ol' b peri-urban Irap sites (triangles) .ind 4 major freshwater wellands (Hags),

ly adjacent to the northern residential suburbs of the city (Whelan 19S7). These are major breeding grounds for mosquitoes and the monitoring and management of mosquitoes is a major health issue for the city (Whelan 1989). Darwin has a monsoonal tropical climate with (.listinct wet and dry seasons (McDonald and McAlpine 1991). The mean annual rainfall is 1,700 mm (Australian Government Bureau of Meteorology 2008). The dry season (x total lainfall = 114 mm) runs from May to October, with a mean maximum temperature of 31.7 C ,ind mean minimum of 21.6 C, and includes 3 Ltmsecutive months (June to August) in which the median monthly rainfall is zero. The wet season runs from November to April and is associated with high relative humidity, intense thunderstorm activity, and monsoonal rains. The mean maxi-

mum temperature during the wet season is 32.3''C and the mean minimum is 24.8 C. In the Darwin area, 3 abundant species of mosquitoes are of particular concern to authorities (Whelan 1989). Cule.x annulirostris and Ae. vigilax are vectors for Ross River and Barmah Forest viruses (Whelan and Weir 1993, Russell 2002, Jacups ct al. 2008). Culex anmdirostris is also the vector for Murray Valley encephalitis and Kunjin viruses (Whelan et al. 1993, Russell and Dwyer 2000). Anopheles farauti s.l. is a potential vector for malaria, a disease that has been eliminated from northern Australia but occurs in neighboring countries to the north (Walker 1998). Anopheles hamroftii and Cq. xanthogastcr arc not disease vectors but are a considerable nuisance to residents (Whelan 1989). All 5 species are widespreati in Australia (Swee-

60

JOURNAL OF THE AMERIC.\N MOSQUITO CONTROL ASSOCIATION

ney et al. 2006, 2007) and most also occur in southeast Asia and/or the Pacific region. Trap and experimental design

Two common types of traps were used: EVS traps baited with 1 kg of dry ice (Rohe and Fall 1979, Cooperband and Cardé 2006) and CDC traps equipped with a clear incandescent 6-V, 3W light bulb (Sudia and Chamberlain 1962). hereafter referred to as EVS CO2 and CDC light traps, respectively. Although EVS COi traps includes a miniature 1.5-V, 70-mA 'grain-ofwheat" light bulb, the main attractant for mosquitoes is COT (Service 1993). Besides using different attractants, these 2 trap types dilTer in size and voltage of their electro-motors. The power sources for CDC light traps used in this experiment were 6-V motorcycle batteries, whereas EVS CO2 traps used 2 1.5-V D batteries. The stronger power source in CDC light traps is needed to supply larger electro-motors and larger-diameter fans (tube diameter ^ 10 cm) plus the light bulbs; tube diameter in the EVS CO2 traps was 7.5 cm with accordingly smaller electro-motors and fans. Sampling was undertaken at 6 sites in the Darwin peri-urban area (Fig. 1). All sites were positioned in mixed woodland close to swamps. Leanycr swamp is a coastal marsh with mangrove vegetation on the boundary, with freshwater, brackish, and saltwater habitats that vary between seasons. Holmes Jungle swamp is also a mixed-character swamp that encompasses a freshwater creek flowing into a freshwater reed swamp, which then empties into a seasonal brackish and saltwater reed swamp whose salinity and mosquito species productivity varies between seasons. Casuarina swamp is based around a freshwater creek within a monsoon forest which Hows into an upper mangrove swamp, whereas Coconut Grove swatnp is based on highly seasonal ephemera! freshwater wetland fed by stormwater drainage, which then leads into a tidal mangrove creek system. Marrara swamp is a seasonal Melaleuea wetland flowing into a relatively long-lasting freshwater Pandanuslfern wetland and freshwater creek. Each site was .sampled with a pair of traps consisting of one of each type placed between 50 m and 100 m apart and separated by a screen of vegetation to minimize interactions between traps. Traps were hung from branches about 1 1.2 m from the ground. Constant locations were maintained for each trap throughout the experiment. One kilogram of dry ice releases CO2 at the rate of 700-750 ml/min (Bishop et al. 2008), which would attract mosquitoes from a range of 2530 m (Gillies 1980), Brown et al. (2008) found that CDC light traps did not recruit beyond 15 m.

Vol.. 2.5. No. 1

Traps were set before sundown and collected the next morning after 7:30 a.m. This was repeated weekly from June 1980 to May 1981, with the exception that dry ice was not available in weeks 2 and 8. Collected traps were stored in a freezer for 2 h to kill the collected insects. The EVS CO2 trap collections were sieved to separate biting midges from mosquitoes, and the CDC light traps were sieved with a series of sieves and then sorted into mosquitoes and non-mosquitoes before identiHcation and counting. Collected mosquitoes were identified by the staff of Medical Entomology in Darwin. Mosquitoes then identified as C'.v. annulirostris are now known to comprise 2 species, most being Cv, annulirostris but a few being Culex palpalis Taylor based on more recent samples at the same sites. Catches of less than 300 were identified and counted fully. Larger catches were subsampled for quantification by identifying and counting samples of ca. 300 specimens, while the primary collection was checked for species absent from the subsample. For catches greater than 300, both the primary collection and the subsample were weighed, with the ratio of weights providing a conversion factor used to estimate the total number of mosquito species in the subsample. The abundance of a species present in the primary coiiection but absent from the subsample was scored as the total number present in the primary collection. Data analysis Abundance data are for females alone. Annual totals have been logio(x + 1) transformed and weekly totals ln(x -1- 1) transformed prior to anaiysis. the more severe transformation being applied to annual totals because of the greater skew in the distribution of species freqtiencies. The 2 wk for which EVS COi data are missing have been excluded from all statistical analyses, though CDC light trap data for these weeks are shown in time series graphs. Using annual totals, we calculated the number of species caught and Simpson's index (1 — D; a measure of diversity ranging from 0 [low diversity] almost to I; Krebs 1989) for each trap. We compared the number of species caught and Simpson's index using matched-pairs /-tests. A comparison of community structures represented by annual totals and proportional abundance of species for each trap was described by ordination using the Bray-Curtis distance measure and nonmetric multidimensional scaling in the software PC-Ord (McCune and Mefford 1999). The 2 species caught in only one or two traps were excluded from the analyses. For each species, we calculated the dilTerence in abundance between paired traps using annual totals and only those sites at which the species

MARni 2009

TRAP TYPE COMPARISON

was present. Mean differences were evaluated using matched-pairs /-tests. We pfesent the associated probabilities r'significanl" ^ P < 0.05) both in raw form and after applying the scqtiential Bonlcrroni correction for multiple lest ing (Rice 1989). Overall trap perlbrmance was evaluated as a single-sample i-test that the mean of species means differed from zero. Kor each of the 5 key species, we examined whether trap typws identified similar seasonal nends in weekly ybundance summed across sites using Pearson correlation coefficients. To evaluiiic whether the seasonality of species inHuenced the ability of trap types to identify similar Irends. we further calculated the correlation between a seasonality index for each species and the abovelnentioned correlation coefficients. The seasonaliiy index employed was the ratio between the highest and lowest running mean for 6 consecutive weeks using EVS CO; data (as these more consistently detected species overall; see Results and Discussion) summed across sites. RESULTS An estimated 93,797 female mosquitoes of 40 s[x;cies were trapped: 19.191 of 37 species in CDC light traps and 74.606 of 37 species in EVS CO2 traps. Individual CDC light traps recorded from 22 to 31 species (x ^ 27.2) and individual EVS COi traps from 27 to 31 species (x ^ 28.8), there being no significant dilTerence between trap tyfies (/